ControlNetAuxSimpleAddText

Don't overthink this one. It stamps a line of text onto an image and gives you the labeled image back. That's the entire node. It's a helper, not a preprocessor - it doesn't detect edges or depth or anything else. In this pack it exists mainly to label preprocessor outputs so you can tell them apart in a side-by-side comparison. If you've seen the pack's "ExecuteAll" example image where every preprocessor result has a caption above it, this is the node that wrote those captions.

What it's for

The practical use is annotation. When you're A/B testing preprocessors - canny vs lineart vs softedge vs depth - you end up with a grid of similar-looking control maps, and it's easy to lose track of which is which. Drop this node after each preprocessor, type the name, and now every map carries its own label. Bake it into a comparison sheet, a preview grid, or a contact sheet you're saving for reference. Outside of that labeling job it's just a generic "put text on an image" utility you can use for whatever you like - watermarks, notes, quick captions.

This is exactly the role it plays in the pack's own testing workflow. The repo ships an "ExecuteAll" example that runs basically every preprocessor at once and tiles the results into a single big comparison image - the kind of sheet you'd use to decide which preprocessor suits a given input. Every tile in that sheet needs a caption or the whole thing is unreadable, and this node is what supplies them. So if you're building your own "run everything, see what sticks" comparison graph - a genuinely good habit when you're learning which preprocessor does what - this is the node that keeps the output legible. That's its whole reason for existing, and it's a reasonable reason.

The inputs and output

Two inputs, one output, nothing hidden:

• image (IMAGE) - the picture to write on.
• text (STRING) - the text to stamp. Type whatever label you want.

The output is an IMAGE - your input with the text drawn onto it. Wire it into a Preview Image, an image-stitching node, or a save node.

Installing it

ComfyUI Manager: search ComfyUI's ControlNet Auxiliary Preprocessors, install, restart. Or manually:

cd ComfyUI/custom_nodes
git clone https://github.com/Fannovel16/comfyui_controlnet_aux
pip install -r requirements.txt

Restart ComfyUI. No model, no download - it works immediately.

Where people get burned

Honestly, there's not much to get wrong here, and if you came looking for a ControlNet feature this isn't one - it's a captioning convenience that happens to ship in the pack. The one thing worth saying: the text it adds becomes part of the image pixels, so if you accidentally leave this node in the path feeding an actual ControlNet, you'll be conditioning on a picture that has words burned into it, which is not what you want. Use it for previews, comparisons, and saved reference sheets - not on the map you're handing to a ControlNet.